chore(dependencies): Bump spinnaker dependencies to 1.0.13 (spinnaker…

…#2409)

fix(dryrun): Rollback cluster stages can run for real

The stage runs a task that identifies the cluster to roll back. That info is useful and it's just a read operation. Also, if that has not happened construction of the after stages that actually perform the rollback operations fails.

feat(core): Pipeline/Orchestration cleanup should only run for redis (s…

…pinnaker#2404)

Optimized variants for sql will be forthcoming.

chore(dependencies): spinnaker-dependencies 1.0.12 (spinnaker#2403)

fix(queue): Move task interceptor execution to within try/catch (spin…

…naker#2399)

Otherwise an exception raised from a failed interceptor will result
in the execution being left in an inconsistent state (`RUNNING`).

Revert "fix(cancel): consistent CANCELED event (spinnaker#2398)" (spi…

…nnaker#2401)

This reverts commit 8a814a1.

fix(core): use correct time interval for agent locks (spinnaker#2400)

fix(cancel): consistent CANCELED event (spinnaker#2398)

Routes cancelation into the CompleteExecutionHandler, allowing it to
determine the final pipeline status, and not updating the pipeline status
until other branches with running stages have been marked as canceled.

chore(*): Bump keiko to 2.9.5 (spinnaker#2396)

fix(clouddriver): stop logging the force cache contents (spinnaker#2397)

It's far too verbose and just clutters the logs in our environment.

If in the future there is interest in improving this, I would suggest
filtering the response down to the cluster being operated on and logging
that.